A method and computer program product are provided for implementing a stable integral equation model for both the time and frequency domains. The stabilization method of the invention is applicable to electric field integral equation (EFIE) formulations, to magnetic field integral equation (MFIE) formulations, and to combined field integral equation (CFIE) formulations. For example, an EFIE formulation can be used in the circuit domain called partial element equivalent circuit (PEEC). Each volume cell is subdivided into a finite number of partitions. A partial inductance term is calculated consisting of a summation over all the finite number of partitions in each cell. Each surface cell is subdivided into a finite number of subcells. A coefficient of potential term is calculated consisting of a summation over all the finite number of subcells in each surface cell. Both the partial inductances and the coefficients of potential are calculated without increasing the number of unknowns. In addition, a damping resistor is added in parallel to the calculated partial inductance in each cell to reduce unwanted resonances outside the useful frequency range.

Patent
   5826215
Priority
Jan 14 1997
Filed
Jan 14 1997
Issued
Oct 20 1998
Expiry
Jan 14 2017
Assg.orig
Entity
Large
18
0
EXPIRED
6. A computer implemented method for creating a stable integral equation, partial Element Equivalent Circuit (PEEC) model for both the time and frequency domains, using electric field integral equation (EFIE) formulations, magnetic field integral equation (MFIE) formulations, or combined field integral equation (CFIE) formulations, or other integration techniques, said computer implemented method comprising the steps of:
subdividing each volume cell into a finite number of partitions;
calculating a partial inductance term consisting of a summation over all the finite number of partitions in each volume cell;
subdividing each surface cell into a finite number of subcells;
calculating a coefficient of potential term consisting of a summation over all the finite number of subcells in each surface cell to create the PEEC model; and
utilizing the created PEEC model for assessing electromagnetic performance.
1. A computer program product used in a computer system having a central processor for creating a stable integral equation, partial Element Equivalent Circuit (PEEC) model for both time and frequency domains using electric field integral equation (EFIE) formulations, magnetic field integral equation (MFIE) formulations, or combined field integral equation (CFIE) formulations, or other integration techniques, the computer program product comprising:
a recording medium;
means, recorded on the recording medium, for subdividing each volume cell into a finite number of partitions;
means, recorded on the recording medium, for calculating a partial inductance term consisting of a summation over all the finite number of partitions in each volume cell;
means, recorded on the recording medium, for subdividing each surface cell into a finite number of subcells;
means, recorded on the recording medium, for calculating a coefficient of potential term consisting of a summation over all the finite number of subcells in each surface cell to create the PEEC model; and
means, recorded on the recording medium, for utilizing the created PEEC model for assessing electromagnetic performance.
2. A computer program product for use in a computer system as recited in claim 1 includes means, recorded on the recording medium, for adding a damping resistor in parallel with a calculated partial inductance term in each cell.
3. A computer program product for use in a computer system as recited in claim 2 includes means, recorded on the recording medium, for using a parallel partial inductance and damping resistor impedance term for loop current calculations.
4. A computer program product for use in a computer system as recited in claim 1 where said means, recorded on the recording medium, for calculating said partial inductance term is represented by ##EQU24## where cell i is partitioned into Mi by Ni sections and cell j is partitioned into Mj by Nj sections; cell i consists of a total of Mi Ni partitions and cell j consists of a total of Mj Nj partitions and the phase term is taken outside of the integral and is approximated by defining R as the distance between the centroid of partition k,kk and of partition q,qq.
5. A computer program product for use in a computer system as recited in claim 1 wherein said means, recorded on the recording medium, for calculating said coefficient of potential term is represented by ##EQU25## for each pair of cells i and j, where each partition has a width w and length l corresponding to the surface of each partition; cell i is partitioned into Mi by Ni sections and cell j is partitioned into Mj by Nj sections; and the phase term is outside of the integral, but is approximated by defining R as the distance between the centroid of partition k,kk and of partition q,qq.
7. A computer implemented method as recited in claim 6 includes the step of adding a damping resistor in parallel with a calculated partial inductance term in each cell.
8. A computer implemented method as recited in claim 7 includes the step of using a parallel partial inductance and damping resistor impedance term for loop current calculations.

The present invention relates to the data processing field, and more particularly, relates to a method and computer program product for implementing a stable integral equation model for both the time and frequency domains. The stabilization method of the invention is applicable to electric field integral equation (EFIE) formulations, to magnetic field integral equation (MFIE) formulations, and to combined field integral equation (CFIE) formulations. An EFIE formulation is used in the circuit domain called partial element equivalent circuit (PEEC). In addition, the stabilization method of the invention is applicable to other numerical integration techniques.

The demand for electromagnetic (EM) modeling capability has increased considerably in the last few years. The modeling of electronic systems for electromagnetic compatibility (EMC) and electrical interconnect and packages (EIP) require efficient solution techniques for large problems which can accommodate a wide range of geometries. These methods should include both radiation as well as scattering formulations. An example of a scattering or incident field problem for electronic systems is the radiated electromagnetic susceptibility to radio and TV signals. These mixed EM and circuits problems are very challenging for electromagnetic formulations both in terms of problem size and complexity.

The Partial Element Equivalent Circuit (PEEC) technique is a circuit based formulation which is numerically equivalent to a full wave method of moments (MoM) solution with Galerkin matching. The Partial Element Equivalent Circuit (PEEC) approach is particularly useful for the modeling of an electromagnetic problem where a circuit interpretation can be used to find more appropriate solutions and where more insight can be gained into the solution process. The PEEC method is based on an Electric Field Integral Equation (EFIE) full wave formulation and it is a circuit domain extension of a Method of Moments (MoM) formulation. The basic PEEC formulation has recently been extended to include incident fields or scattering (sPEEC), dielectrics, lossy dielectrics, and the use of the modified nodal as well as the modified loop analysis (MLA) formulation.

The applicability of the PEEC model to both time and frequency domain problems is very useful. Very often, additional insight can be found from the analysis of a problem in both domains. Simple PEEC models which do not involve delays can be applied to both domains using a conventional circuit solver computer program, like SPICE or ASTAP. Instabilities for the time domain integral equation formulations for EM problems have been observed by many authors. Many different techniques have been suggested to overcome these instabilities including a conjugate gradient method, a filtering technique, and a matrix pencil technique to eliminate late time instabilities. While all of these techniques help eliminate some of stability problems for specific cases, some sources of instabilities remain.

A recent paper on stability by A. E. Ruehli et al., is entitled "Stability of Discretized Partial Element Equivalent EFIE Circuit Models", IEEE Transactions on Antennas and Propagation, 43(6):553-559, June 1995. In this paper, a potential stabilization scheme for the time domain was proposed using the PEEC interpretation of the integral equation. Each inductive self cell was subdivided into two series inductances and a delay τs was introduced between these two inductances. It was shown analytically for a small model how the roots of the system could be moved from the right half plane to the left half plane by the proper assignment of τs. In this scheme the delays were adjusted to obtain a stable model. It was observed that the discretization of the continuous time EFIE leads to an unstable discrete system since the model has poles in the right half plane. For more complex problems, this stabilization scheme was insufficient.

It is known from circuit theory that the PEEC models without delays are passive and that they are always stable. Hence, it is evident that the instabilities are caused by the delays between the partial inductances and the coefficients of potential. An interesting fact is that well implemented MoM and PEEC formulations with delay do give the same response in the region of interest where the cells are chosen to be sufficiently small such that the cell size is at least 20 cells/λ. Note that impedance computations, like the ones considered in this specification, do require more cells for an accurate solution than do radiation problems.

In the present specification, an active frequency range is defined as a lower frequency range corresponding to a wavelength where all cells are of a size ≧20 cells/λ. In models investigated for much higher frequencies corresponding to ≦20 cells/λ, this frequency range is defined as an extended frequency range. The motivation for computations in the extended frequency range is the fact that a complex resonance behavior of the models are observed in this range. This may be totally irrelevant for a straight forward frequency analysis but is extremely important for other applications like the time domain. A need exists to eliminate the resonance behavior in the extended frequency range.

A principal object of the present invention is to provide an improved method and computer program product for implementing integral equation models using partial element equivalent circuit (PEEC). Other important objects of the present invention are to provide such method and computer program product substantially without negative effects and that overcome some disadvantages of prior art arrangements.

In brief, a method and computer program product are provided for implementing a stable integral equation model for both the time and frequency domains. The stabilization method of the invention is applicable to electric field integral equation (EFIE) formulations, to magnetic field integral equation (MFIE) formulations, and to combined field integral equation (CFIE) formulations. For example, an EFIE formulation can be used in the circuit domain called partial element equivalent circuit (PEEC). Each volume cell is subdivided into a finite number of partitions. A partial inductance term is calculated consisting of a summation over all the finite number of partitions in each cell. Each surface cell is subdivided into a finite number of subcells. A coefficient of potential term is calculated consisting of a summation over all the finite number of subcells in each surface cell.

In accordance with features of the invention both the partial inductances and the coefficients of potential are calculated without increasing the number of unknowns. In addition, a damping resistor is added in parallel to the calculated partial inductance in each cell to reduce unwanted resonances outside the useful frequency range.

The present invention together with the above and other objects and advantages may best be understood from the following detailed description of the preferred embodiments of the invention illustrated in the drawings, wherein:

FIG. 1 is a block diagram representation of a computer or data processing system of the preferred embodiment;

FIG. 2 is a diagram illustrating a conductor example that has been discretized in a PEEC formulation into two volume cells and three surface cells;

FIG. 3 is a diagram illustrating a PEEC model for the conductor example of FIG. 2 of the preferred embodiment;

FIG. 4 is a diagram illustrating a PEEC model including damping resistors for the conductor example of FIG. 2 of the preferred embodiment;

FIGS. 5A, 5B, and 5C illustrate three different geometries including a patch antenna and two different folded dipole antennas for computer experiments of the preferred embodiment;

FIGS. 6 through 15 are charts illustrating results of computer experiments of the preferred embodiment for the three different geometries including a patch antenna and two different folded dipole antennas;

FIG. 16 is a logic flowchart illustrating the method of the preferred embodiment; and

FIG. 17 is a block diagram illustrating a computer program product in accordance with the preferred embodiment.

Having reference now to the drawings, in FIG. 1 there is shown a computer or data processing system of the preferred embodiment generally designated by the reference character 100. As shown in FIG. 1, computer system 100 includes a central processor unit (CPU) 101, a read only memory 102, a random access memory 104, a display adapter 106 coupled to a display 108. CPU 101 is connected to a user interface (UI) adapter 110 connected to a pointer device and keyboard 112. CPU 101 is connected to an input/output (IO) adapter 114 connected to a direct access storage device (DASD) 116 and a tape unit 118. CPU 101 is connected to a communications adapter 120 providing a communications function.

Various commercially available processors could be used for computer system 100, for example, an IBM personal computer or similar workstation can be used. Central processor unit 101 is suitably programmed to execute the logic flowchart of FIG. 16 of the preferred embodiment.

In accordance with features of the invention, improvements are made to the PEEC model through the development of a refined method of computing both the partial inductances as well as the coefficients of potential without increasing the number of unknowns. The improved PEEC model of the present invention is referenced as +PEEC. In addition, damping is added to the PEEC model in order to reduce unwanted resonances outside the useful frequency range. The improved PEEC model of the present invention with added damping is referenced as R+PEEC. The effectiveness of the method is illustrated with several examples illustrated and described with respect to FIGS. 6-15. Also, in FIG. 6 the results of the present invention are compared with a conventional Method of Moments (MoM) code.

First a derivation of the PEEC models for both partial inductance and for coefficients of potential beginning with the electric field integral equation (EFIE). Next a derivation of the PEEC model improvements of the present invention is provided (+PEEC) and (R+PEEC).

Now an appropriate integral equation formulation is derived for the PEEC model. Then from this integral equation formulation both the partial inductance and the coefficient of potential terms are defined for the conventional PEEC model. The partial inductance and coefficient of potential terms defined for the conventional PEEC model provide the foundation for the improvements of the +PEEC model of the preferred embodiment.

The electric field in free space at a conductor can be written as ##EQU1## where J is the current density in the conductor, q is the charge density, and G is the free space Green's function ##EQU2## where R is given by R=|r-r'|.

At the surface of a perfect conductor, ##EQU3## and the tangential component of the electric field E=0. This boundary condition is applied to the surfaces of the perfect conductor in Eq. 1 resulting in the following electric field integral equation ##EQU4## where n is a vector normal to the surface of the conductor, and SPEC is surface of a perfect electrical conductor.

The unknown quantities J and q are assumed constant over a cell. The geometry should be discretized into small enough cells so that this assumption is valid. This is less of an issue for a full wave solution since the cells must be much smaller than the wavelength. In the PEEC formulation, both q and J are unknowns and are related by the continuity equation ∇·J+jωq=0. The continuity equation is enforced by writing Kirchoff's current law (KCL) at each node. The vector quantities are discretized into orthogonal coordinates as J=Jx x+Jy y+Jz z. Substituting this relationship into Eq. 1 will result in a set of three vector equations which are identical in form, but differ in spatial directions x, y, z. As an example, consider the current for the conductor shown in FIG. 2 which has been discretized into 2 volume cells α and β and 3 surface cells γ, δ, ζ.

Substitution of the z component of the current into the integral equation for this example results in ##EQU5##

As in any method of moments formulation, basis and testing functions are selected. In the PEEC formulation, there are two unknowns, current density J and charge q. Each unknown is expanded into a series of pulse basis functions with an unknown amplitude. Pulse functions are also selected for the testing functions for the Galerkin formulation at hand. The cell to cell coupling is defined with respect to cell α in FIG. 2 by integrating each term of the integral equation 3 over the volume vα of conductor cell α in the following manner. ##EQU6## where vα is the volume of the conductor cell α, aα is the cross-section of the cell (x-y direction) and lα is the length (z-direction) and f(r) is the integrand. We can view this either as the Galerkin solution or as averaging the field at point r over the cell volume vα.

Starting with the integral equation 4, the first 2 terms, or vector potential terms, are shown to correspond to partial inductances. Keeping in mind that the current density J is constant over each cell and Iz =Jz αcell, where αcell is the cross sectional area of a cell, equation 4 becomes ##EQU7## Now, by applying the equation 5, the vector potential terms in equation 6 become ##EQU8## The partial mutual inductance between cells α and β is defined as ##EQU9## Using this definition and equation 2 for the Green's Function, equation 7 can be rewritten as

jωLpαα Iα +jωLpαβ Iβ e-jβR αβ (Eq. 9)

where the first term is the partial self inductance of the cell α and the second term represents the inductive coupling to cell α from a current in cell β in FIG. 2. The delay between cells α and β are given by the phase term e-jβR αβ where Rαβ is the center to center distance between cells α and β. In general, given n cells, the inductive coupling terms is ##EQU10##

In a similar manner, the coefficient of potential terms can be derived from the last 3 terms in equation 4. With the charge q is constant over each equation 4 becomes ##EQU11## Now, by applying Eq. 5, we get ##EQU12## The surface integral can be redefined at

F(z)=∫S G(r,r')dS (Eq. 13)

With this, each term in equation 12 can be rewritten in terms of the F(z) ##EQU13## Each volume integral can be broken into 2 integrals ##EQU14## The central difference approximation may be used to evaluate the derivative. ##EQU15## where ##EQU16## and similarly ##EQU17## With this result, equation 11 becomes ##EQU18## The definition of coefficient of potential is given in as ##EQU19##

The total charge on cell γ is given by Qγ =qγ aγ where aγ is the surface area of cell γ. With this relationship and the coefficient of potential definition in Eq. 18, each term in Eq. 17 can be rewritten. The delay between cell α and cells γ, δ, and ζ are given by the phase terms.

Qγ Pγγ +Qδ Pγδ e-jβRγδ +Qζ Pγ ζe-jβRγζ (Eq. 19)

where Rγδ is the center to center distance between cells γ and δ.

Historically, PEEC models have been implemented using partial inductances and coefficients of potential as they are defined above in equations 8 and 18. As can be seen from these definitions, they are based on analytically integrating the 1/R dependence of the Green's function over the appropriate cells. The phase term of the Green's function is not included under the integral in order to be compatible with the time domain. Instead, the integration of the phase term is approximated by the evaluation at the centroid of each cell. This approximation provides very good answers for the frequency domain in comparison with other MoM methods. In addition, numerical rather than analytical integration would have been necessary if the phase terms had been included in equations 8 and 18 for the Galerkin solution.

Now a new way of calculating the partial inductance and coefficient of potential terms is derived in accordance with the present invention which also can be implemented both in the time as well as the frequency domain.

The partial inductance and coefficient of the potential terms have been derived from the PEEC interpretation of the EFIE. Next, in the enhanced PEEC model of the invention is developed which is called +PEEC, for notational convenience.

The modified loop analysis (MLA) method is used to implement both the PEEC and +PEEC formulations. Using the MLA method, the PEEC model for the 2 cell conductor example of FIG. 2 is shown in FIG. 3. For this geometry, the MLA method is the most efficient implementation for PEEC in the frequency domain and it corresponds to the usual MoM impedance matrix for simple cases.

It is well known that for impedance problems, about 20 cells/λ at ƒmax is required for an accurate answer up to some maximum frequency ƒmax. It is clear that we could obtain a clean frequency response up to 50 ƒmax by decreasing the cell size by about a factor 50 in both directions, but this increased range comes at the high price of an extremely large number of unknowns. In all our examples we chose ƒmax =20 GHz and we tried to obtain a clean model for the extended frequency range. The feature of +PEEC is that the behavior of the solution for frequencies in the extended frequency range is stabilized without increasing the number of unknowns. To achieve this feature, two changes were implemented in PEEC which are applied simultaneously to obtain the +PEEC model. These changes include improved partial inductances and coefficients of potential. Along with these model improvements, a third change is implemented through the introduction of damping resistors. This combined model with all three improvements is referred to as R+PEEC. Now the implementation of +PEEC and R+PEEC are described and results are illustrated in FIGS. 6-15 which show that +PEEC and R+PEEC stabilizes the PEEC models.

+PEEC and R+PEEC increase in accuracy of the phase behavior for the extended frequency range without introducing more unknowns. Having reference to the 2 cell example of FIGS. 2 and 3, each cell can be considered to be at the macromodel level. By subdividing each of these cells into a finite number of partitions, a new way of computing the inductance term for each cell at the macromodel level is introduced. By using the relationship in Eq. 10 for this 2 cell equation provides

jωLp11 I1 +jωLp12 I2 ie-jβR 12

jωLp22 I2 +jωLp21 I1 ie-jβR 21 (Eq. 20)

We can now subdivide each of these cells into a finite number of partitions given by

Δ=c/nƒe

WHERE c is the speed of light in free space, n=1, 2, . . . , and ƒe is given by the upper end of the extended frequency range. In our example, we used ƒe =1000 GHz. By allowing n to vary, the smallest number of partitions for each geometry can be selected such that the problem is stable, and the computational effort is minimized. Importantly, the unknown current for each cell is not further subdivided resulting in the same number of unknowns. This partitioning of each cell results in each partial inductance term consisting of a summation over all the partitions in each cell. For each cell, equation 8 becomes ##EQU20## where cell i is partitioned into Mi by Ni sections and cell j is partitioned into Mj by Nj sections. Cell i consists of a total of Mi Ni partitions and cell j consists of a total of Mj Nj partitions. The phase term is taken outside of the integral and is approximated by defining R as the distance between the centroid of partition k,kk and of partition q,qq.

Each partition now is treated the same as each cell. For example, where there was previously one calculation for a partial self inductance for cell i, now there are 4 summations over all Mi Ni partitions contained within cell i. For the mutual terms, the calculation is similar in that there are 4 summations, where 2 summations are over the partitions due to cell i and 2 summations are over the partitions due to cell j as is shown in equation 21 of the preferred embodiment.

The computation of the partial self and partial mutual inductance terms has increased to the extent that Mi Ni operations are required for each partial inductance term. However, the analytical integration is computed only for the self cell terms. For all mutual terms, the integral is approximated by 1/R using the centroid to centroid distance of each partition within each cell. Although the computation of the partial inductance terms has changed considerably with this scheme, the unknowns are the same as before.

In a similar manner, the coefficient of potential cells γ, δ, and ζ in FIG. 2 is partitioned into subcells. In general, for each pair of cells i and j, the coefficient of potential calculation with the delay term is replaced by ##EQU21## where each partition has a width w and length l corresponding to the surface of each partition. Cell i is partitioned into Mi by Ni sections and cell j is partitioned into Mj by Nj sections. The phase term is again outside of the integral, but is approximated by defining R as the distance between the centroid of partition k,kk and of partition q,qq.

As before, the total number of PEEC capacitance models has not changed, but just the actual coefficient of potential calculation has changed. Therefore, the number of unknowns in the problem has not changed.

For computer experiments, three different geometries were used as shown in FIGS. 5A, 5B, and 5C including a patch antenna and two folded dipole antennas. These are quite different problems since the patch antenna is capacitive and the folded dipole antennas are inductive at low frequencies. A problem is unstable if the real portion of the input impedance is shown to be negative. By implementing the calculation of both the partial inductance and coefficients of potential of the preferred embodiment, the input impedance is positive for a much larger frequency range. In addition to this improvements provided by the +PEEC method, the useful frequency range is further extended, without introducing more unknowns, by the addition of a damping resistor Rαα, Rββ in parallel with the partial inductance, as shown in FIG. 4. The PEEC and +PEEC methods are powerful because circuit techniques can be used to interpret and analyze problems.

With the added damping resistor in parallel to the partial inductances, the R+PEEC model is provided. Since each loop current through the partial inductance is an unknown, by adding the parallel resistor, one additional loop current is added for each existing unknown. The challenge is to implement this parallel resistance without changing the size of the matrix, i.e., the number of unknowns. The unknown loop current in cell α is given by Iα. This loop current can be written in terms of the currents through the inductor and resistor in loop α as Iα =IRαα +ILpαα. The parallel impedance is given by ##EQU22## By using this parallel impedance, we do not introduce an additional unknown into the MLA matrix. In general, the stamp would be of the following form for a given current ILk going through the kk-th partial inductance.

______________________________________
Row . . . ILk . . . RHS
______________________________________
k 0
##STR1## 0 0
______________________________________

The stamp would be of the following form for the coupling of the current ILn into each of the other partial inductance terms.

______________________________________
Row . . . ILn . . . RHS
______________________________________
k . . .
##STR2## 0 0
______________________________________
where
##STR3##
- The actual value of the damping resistor is calculated for each cell
using the following equation

Rii =kωmax Lpii (Eq. 24)

where ωmax =2πƒmax, and k=10, . . . 100, depending on the problem. Using this equation for R, k may be adjusted such that R dominates at frequencies greater than ƒmax, but not at frequencies lower than ƒmax.

The +PEEC and the R+PEEC improvements to the basic PEEC model in Section 3 are result of extensive computer experiments. Also, the fact that the techniques must be applicable both in the frequency as well as the time domain is a key constraint in accordance with the preferred embodiment. Having reference to FIGS. 6-15, experimental evidence of the benefits of the +PEEC and the R+PEEC model improvements is provided. We discretize each example for 20 cells/λ for the active frequency range. For each example, the active frequency range is given by ƒmax =20 GHz, and the extended frequency range is given by ƒe =1000 GHz.

In FIG. 6 the results of the of the +PEEC model is compared with a carefully implemented MoM code. The first comparison is for a patch antenna where the MoM and +PEEC is compared for the active frequency range. The patch antenna is center fed and has a length of 9 mm and a width of 4.5 mm, and is shown in FIG. 5A. The real and imaginary input impedances are shown in FIG. 6 and are in close agreement. The second comparison for the active frequency region is shown in FIG. 7. Again, the same patch antenna is used for a comparison between a basic PEEC model and +PEEC model. In FIG. 7, the same close agreement is shown. Based on these and other examples, both PEEC and +PEEC agree very closely with MoM results in the active frequency range.

In order to show the effects of the improvements of +PEEC, the input impedance for the same patch antenna at 20 cells/λ at 20 GHz for the extended region is shown. This data is shown in FIGS. 8 and 9 for the real and imaginary input impedances for the patch antenna. This partitioning scheme (+PEEC) has significantly reduced the spurious resonances throughout the extended region.

The frequency response above ƒmax is quite surprising for both MoM and PEEC techniques. FIG. 10 shows a comparison of two different discretizations, one is properly discretized for the active region, the other has a few additional cells. This data clearly shows the discretization dependence of the resonance behavior. In addition, these results clearly explain why these models lead to late time instabilities in the time domain.

Having reference to FIG. 11, the effects of varying the partitioning scheme in +PEEC are illustrated. The partitioning is varied using n in ##EQU23## In the previous results, n=3 was used. FIG. 11 shows the variation in the real input impedance for n=1, 2, 3. This data clearly illustrates the power of this partitioning technique. For n=1, the real input impedance is positive up to 450 GHz, for n=2 it is positive up to 850 GHz, and for n=3 it is positive up to 1000 GHz. In addition, the negative spike at 100-150 GHz can be eliminated by increasing the number of partitions used.

Next, the impact of the damping resistor is compared (R+PEEC) with the +PEEC model using n=3 partitions. These results are shown for the real input impedance in FIG. 12. It is clear that the R reduces the spike occurring between 100-150 GHz for all selected k values.

In summary, the results for n=3 both with and without a resistor are shown in FIG. 13. It is clear that the results are stable for the entire extended frequency range for R+PEEC when n=3, k=100. Referring again to FIG. 8, we see the remarkable difference between these results and the original PEEC results before the stabilization of +PEEC of the preferred embodiment.

The first folded dipole to be examined is referred to as a loop as shown in FIG. 5B and the loop has an edge to edge length of 9.9 mm, and the loop has an edge to edge width of 5.4 mm. Each flat conductor has a width of 0.9 mm. The results are shown in FIG. 14 for PEEC and R+PEEC (n=3, k=50). It is clear the stabilization scheme has eliminated the false resonances throughout the extended frequency range.

The second folded dipole to be examined is referred to as a ribbon folded dipole as shown in FIG. 5C. The ribbon has an edge to edge length of 9 mm, and an edge to edge width of 4.5 mm. Each flat conductor has a width of 0.9 mm. The number of partitions was selected as n=4 and k=10 was used for the resistor value. A comparison of PEEC, +PEEC, and R+PEEC for the real part of the input impedance is shown in FIG. 15. The results show that the partitioning scheme (+PEEC) provides a stable solution, while the damping resistor (R+PEEC) provides additional smoothing over the extended frequency range.

The false resonances observed explain why the late time instabilities occur for integral equation type models. The process of eliminating these resonances is quite challenging. The solution of the +PEEC, R+PEEC models is assisted by the fact that the PEEC techniques transform the problem into the circuit domain. This allows the use circuit techniques to introduce the damping resistance without increasing the number of unknowns.

In summary, the solution of the preferred embodiment is twofold, implementing the partitioning scheme for the partial inductances and coefficients of potential, and introducing a damping resistor in parallel with each partial inductance. Both the partitioning scheme and the damping resistor are developed in a general sense, such that they are applicable to any geometry. In addition, the parallel resistance may be used to model losses that occur in a real problem. It is observed that many formulations lead to negative input impedances above ƒmax unless the geometry is further discretized into more cells. Increasing the discretization is not very desirable since the number of unknowns is increased drastically which directly impacts the solution time and potentially the size of the problem which can be solved. Also, it is important to observe that the +PEEC improvement techniques of the preferred embodiment are applicable to the time domain formulation as well as to the frequency domain.

Referring to FIG. 16, there are shown exemplary steps according to the teachings of the invention for implementing integral equation models using partial element equivalent circuit (+PEEC) of the preferred embodiment starting at block 1600. First the variable n is selected and each of the cells in a PEEC model are subdivided into a finite number of partitions as indicated at a block 1602. A partial inductance is calculated consisting of a summation over all the finite number of partitions in each cell as indicated at a block 1604. Each surface cell is suddivided into a finite number of subcells as indicated at a block 1606. A coefficient of potential term is calculated consisting of a summation over the finite number of subcells in each surface cell as indicated at a block 1608. Next a damping resistor is added in paralled with the calculated partial inductance term in each cell as indicated at a block 1610.

Referring now to FIG. 17, an article of manufacture or a computer program product 1700 of the invention is illustrated. The computer program product 1700 includes a recording medium 1702, such as, a floppy disk, a high capacity read only memory in the form of an optically read compact disk or CD-ROM, a tape, a transmission type media such as a digital or analog communications link, or a similar computer program product. Recording medium 1702 stores program means 1704, 1706, 1708, 1710, on the medium 1702 for carrying out the methods of the preferred embodiment in the system 100 of FIG. 1.

A sequence of program instructions or a logical assembly of one or more interrelated modules defined by the recorded program means 1704, 1706, 1708, 1710, direct the computer system 100 for implementing integral equation models using partial element equivalent circuit (+PEEC and R+PEEC) of the preferred embodiment.

While the present invention has been described with reference to the details of the embodiments of the invention shown in the drawing, these details are not intended to limit the scope of the invention as claimed in the appended claims.

Ruehli, Albert Emil, Garrett, Jan Elizabeth

Patent Priority Assignee Title
6058258, Oct 28 1997 GOOGLE LLC Method for analyzing the stability and passivity of system models
6188974, Mar 23 1998 IBM Corporation Method for constructing reduced-order models of systems
6606586, Nov 27 1998 Mitsubishi Denki Kabushiki Kaisha Simulation method
7310595, Aug 19 2002 Intersil Americas Inc. Numerically modeling inductive circuit elements
7496871, Oct 21 2003 Siemens Industry Software Inc Mutual inductance extraction using dipole approximations
7546230, Apr 18 2005 GLOBALFOUNDRIES Inc Electromagnetic response model with improved high frequency stability
7707527, Sep 27 2007 International Business Machines Corporation; University of Illinois at Urbana—Champaign Huygens' box methodology for signal integrity analysis
8065129, Nov 19 2004 SYNOPSYS, INC , A DELAWARE CORPORATION Methods and apparatuses for circuit simulation
8146043, Sep 27 2007 International Business Machines Corporation Huygens' box methodology for signal integrity analysis
8161438, Oct 21 2003 Siemens Industry Software Inc Determining mutual inductance between intentional inductors
8214788, Mar 08 2008 Siemens Industry Software Inc High-frequency VLSI interconnect and intentional inductor impedance extraction in the presence of a multi-layer conductive substrate
8249849, May 29 2008 Fujitsu Limited Power source network analyzing apparatus, power source network analyzing method, and power source network analyzing program
8549449, Oct 21 2003 Siemens Industry Software Inc Mutual inductance extraction using dipole approximations
8650522, Oct 21 2003 Siemens Industry Software Inc Determining mutual inductance between intentional inductors
8732648, Mar 08 2008 Siemens Industry Software Inc High-frequency VLSI interconnect and intentional inductor impedance extraction in the presence of a multi-layer conductive substrate
8826204, Oct 21 2003 Siemens Industry Software Inc Mutual inductance extraction using dipole approximations
8910108, Mar 08 2008 Siemens Industry Software Inc High-frequency VLSI interconnect and intentional inductor impedance extraction in the presence of a multi-layer conductive substrate
9230054, Mar 08 2008 Siemens Industry Software Inc High-frequency VLSI interconnect and intentional inductor impedance extraction in the presence of a multi-layer conductive substrate
Patent Priority Assignee Title
///
Executed onAssignorAssigneeConveyanceFrameReelDoc
Jan 13 1997GARRETT, JAN ELIZABETHInternational Business Machines CorporationASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS 0086220155 pdf
Jan 14 1997International Business Machines Corporation(assignment on the face of the patent)
Feb 14 1997RUEHLI, ALBERT EMILInternational Business Machines CorporationASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS 0086220155 pdf
Date Maintenance Fee Events
May 07 2002REM: Maintenance Fee Reminder Mailed.
Oct 21 2002EXP: Patent Expired for Failure to Pay Maintenance Fees.
Nov 20 2002EXP: Patent Expired for Failure to Pay Maintenance Fees.


Date Maintenance Schedule
Oct 20 20014 years fee payment window open
Apr 20 20026 months grace period start (w surcharge)
Oct 20 2002patent expiry (for year 4)
Oct 20 20042 years to revive unintentionally abandoned end. (for year 4)
Oct 20 20058 years fee payment window open
Apr 20 20066 months grace period start (w surcharge)
Oct 20 2006patent expiry (for year 8)
Oct 20 20082 years to revive unintentionally abandoned end. (for year 8)
Oct 20 200912 years fee payment window open
Apr 20 20106 months grace period start (w surcharge)
Oct 20 2010patent expiry (for year 12)
Oct 20 20122 years to revive unintentionally abandoned end. (for year 12)